Scientometrics (2017) 111:1385–1413 DOI 10.1007/s11192-017-2350-x The impact of individual collaborative activities on knowledge creation and transmission 1,2 1,3 Nuha Zamzami • Andrea Schiffauerova Received: 18 May 2016 / Published online: 18 March 2017 Ó Akade´miai Kiado´, Budapest, Hungary 2017 Abstract Collaboration is a major factor in the knowledge and innovation creation in emerging science-driven industries where the technology is rapidly changing and con- stantly evolving, such as nanotechnology. The objective of this work is to investigate the role of individual scientists and their collaborations in enhancing the knowledge flows, and consequently the scientific production. The methodology involves two main phases. First, the data on all the nanotechnology journal publications in Canada was extracted from the SCOPUS database to create the co-authorship network, and then employ statistical data mining techniques to analyze the scientists’ research performance and partnership history. Also, a questionnaire was sent directly to the researchers selected from our database seeking the predominant properties that make a scientist sufficiently attractive to be selected as a research partner. In the second phase, an agent-based model using Netlogo has been developed to study the network in its dynamic context where several factors could be controlled. It was found that scientists in centralized positions in such networks have a considerable positive impact on the knowledge flows, while loyalty and strong connections within a dense local research group negatively affect the knowledge transmission. Star scientists appear to play a substitutive role in the network and are selected when the usual collaborators, i.e., most famous, and trustable partners are scarce or missing. Keywords Scientific collaboration Á Partnership Á Productivity Á Knowledge flows Á Social network Á Network structure & Nuha Zamzami [email protected] 1 Concordia Institute for Information Systems Engineering, Concordia University, Montreal, QC, Canada 2 Faculty of Computing and Information Technology, King Abdul-Aziz University, Jeddah, Saudi Arabia 3 Department of Engineering Systems and Management, Masdar Institute of Science and Technology, Abu Dhabi, UAE 123 1386 Scientometrics (2017) 111:1385–1413 Introduction and overview In today’s rapidly growing technological fields the sources of knowledge are widely dis- tributed. Solving new rising issues and answering many complex and multidisciplinary research questions requires higher level of skills and comprehensive knowledge. This leads to the need for collaborative knowledge sharing whose ability to address more complex and critical research problems has already been demonstrated in the literature (Sonnenwald 2007). Moreover, a significant increase in research productivity as result of scientific collaboration has been suggested by several scholars (e.g. Price and Beaver 1966; Zuck- erman 1967; Gla¨nzel and Winterhager 1992; Landry et al. 1996) The collaborative activities can be mapped as a complex network, where its nodes represent the collaborators and their partnerships form the links. In such networks, the knowledge is created and transmitted by socially connected individuals whose collabora- tions shape the links of the networks. In this work, it is the individual researchers who are the network nodes and their co-authorship of journal articles are the partnership linkages between these nodes, i.e., partners co-creating knowledge through their scientific collaboration. The knowledge creation network is a dynamic social network where the behavior of collaborators is influenced by their interactions with others over an interval of time. Scholars have analyzed the performance of such networks in the last decades in order to derive policy implications, and to enhance the institutional and governmental decision- making in the area of innovation policy. The existing research studies based on the dynamic social networks approach mainly focused on the firm level analysis, for example analyzing different categories of firms, such as firm leaders or startups (Nagpaul 2002), or studying the roles of various institutions such as academic, industry and governmental ones (Scholz et al. 2010; Triulzi et al. 2011), but much less of the research has been carried out at individual level (Wang 2013; Tajaddod Alizadeh et al. 2015). As the overall productivity of the network depends on the performance of its actors, the quantity and speed of knowledge diffusion is greatly affected by the individual collabo- rative activities (Pyka et al. 2002). That is, individuals with certain characteristics would facilitate the network’s productivity while the behavior of others might have more negative effects on the knowledge creation and transmission within the network. For example, star scientists, researchers with high impact on innovation and knowledge development reflected by their considerably higher productivity comparing to their colleagues and competitors, are important in the process of technology transfer. The importance of star scientists is not only because of their outstanding scientific knowledge and performance that contribute significantly to the success of firms (Zucker and Darby 2005) but also because they actually act as knowledge circulation improvers in the scientific networks (Schiffauerova and Beaudry 2012). Moreover, gatekeepers, the nodes with highest betweenness centrality, are the influential individuals who are responsible for the knowl- edge transfer in the network where they interconnect different individuals or bridge sep- arate research groups, and help this way in merging various existing ideas (Gould and Fernandez 1989). Furthermore, popular researchers, those who are connected to a greater number of collaborators, are critically important for sharing the knowledge considering their ability to access a significant amount of fresh and new knowledge and to bring that knowledge to their colleagues for further collaborative activities which will positively affect the pro- ductivity (Henderson and Cockburn 1996). Besides, a better performance for scientist can 123 Scientometrics (2017) 111:1385–1413 1387 be detected in the case of maintaining strong ties with previous partners, i.e. loyal sci- entists, rather than having several co-authorship relationships with multiple ones (Abbasi and Altmann 2011). On the other hand, scientists who are much willing to collaborate with the neighbors of their neighbors, i.e. embedded scientists, are more likely to be deeply involved in a local network of collaboration (his/her own research group) which will discourage the knowledge transmission within the network (Breschi and Lissoni 2006). The presence of star scientists, gatekeepers, loyal scientists or popular scientists seems to have an impact on the knowledge flows within the network and, consequently, it can be assumed that they also influence the scientific productivity of the other network actors and of the network as a whole. This work therefore aims to first analyze and understand the collaborative knowledge sharing behavior in the real world, and then to simulate the Canadian nanotechnology knowledge-based network and evaluate its dynamics. Our concern is to study the impact of specific groups of researchers on the collaborative behavior of others within the network as well as on the structure, productivity and effi- ciency of the whole network. We have created the network at individual level based on the co-authorship relationships between Canadian scientists publishing in the field of nan- otechnology. The network is then studied in a dynamic context using an agent-based modeling approach. The contribution of this work is twofold: First; It provides a detailed examination of the distinguished individual scientists’ roles which invloves the study of their impact on the productivity, on the network structure and also on the perfromance of other sceintists in the network. Second; There is also a methodological contribution, because this work takes a dynamic perspective while basing the whole study on the real data on the scientists collected by the authors using various methods (such as statistical analysis, data mining and survey). The paper is organized as follows. ‘‘Review of the literature’’ section reviews the literature on the networks of collaborators and their research performance. ‘‘Research methods’’ section describes the data upon which our analyses are based and presents the methodology used. ‘‘Results’’ section discusses the analysis performed and reports the results. ‘‘Discussion and concluding remarks’’ section concludes the findings and proposes some new research opportunities. Review of the literature Scientific performance of individuals Several studies on collaboration suggested significant increase in research productivity related to the collaborative activity (e.g. Price and Beaver 1966; Zuckerman 1967; Gla¨nzel and Winterhager 1992; Landry et al. 1996). Analyzing scientific papers over an interval of time while considering the percentage of articles written by co-authors showed a positive relationship between collaboration and higher productivity, which shows that collaborative knowledge creation played an essential positive role in the scientific performance (Beaver and Rosen 1979; Allen 1983; Drejer and Vinding 2006; Manley et al. 2009). Most of the scientific output is typically produced by the top 1% or 2% of scientists working in a specific area. These most productive scientists are generally called star